This proposal outlines a plan of research to better understand the causes of racial residential segregation in American cities. Racial residential segregation contributes to the formation of high-poverty neighborhoods and has been implicated as an important source of enduring racial health disparities (e.g., Mechanic 2007). Yet social scientists do not know what forces contribute to maintaining residential racial segregation, what the respective importance of these forces are, or how these forces combine. Two obstacles limit accumulation of causal knowledge in this research area. First, we lack a plausible model of how individual demographic factors and neighborhood characteristics simultaneously affect individual mobility decisions. Second, most past research measures potential explanatory factors at the individual or intermediary (e.g., real-estate agents or lenders) level, but do not represent how those factors affect neighborhood racial composition in the aggregate. Our project addresses these problems through a discrete choice statistical model of residential choice and an agent-based simulation of neighborhood formation. The discrete choice models are estimated using the data on residential mobility from the Panel Study of Income Dynamics matched with data from the Decennial Censuses. Our basic model of residential mobility incorporates race and neighborhood racial composition, income and neighborhood housing cost, neighborhood income composition, housing tenure (owning or renting), and household composition;proposed extensions to the model incorporate family wealth, mobility impacts of nearby neighborhood changes, and housing market discrimination. The agent-based model creates simulated cities with the demography, geography, and housing stock of actual urban areas in which household mobility is governed by behavioral rules estimated in the discrete choice model. Taken together, this framework can be used to address questions regarding how preferences for race of neighbors, affordability constraints, housing market discrimination, and metropolitan population composition affect racial segregation. The model can also be used to explore the effects of spatially targeted housing policies on residential segregation, like the demolition of distressed public housing under Hope VI legislation. PUBLIC HEALTH RELEVANCE: Racial residential segregation contributes to the formation of high-poverty neighborhoods and has been implicated as an important source of enduring racial health disparities (e.g., Mechanic 2007). One strategy for reducing health disparities among racial groups, particularly among blacks and whites, would be to reduce racial residential segregation and the corresponding geographic concentration of poverty. The proposed research aims to provide a better understanding of the causes of segregation, suggest solutions for reducing neighborhood racial segregation, and develop new analytical and technical tools for modeling dynamic processes in scientific research.